This qwen2 model was trained 2x faster with Unsloth and Huggingface's TRL library.
unsloth/Qwen2.5-7B-Instruct
